The technological background is illustrated by patent applications WO-A-90.12851, WO-A-96.20255, EP-A-0 036 151, EP-A-0 036 609, EP-A-0 272 378, and FR-A-2 647 804.
The steam-cracking process is the basic process of the petrochemical industry and consists in cracking a feedstock of hydrocarbons and water vapor at high temperature and then abruptly cooling it. The main operating problem arises from the deposition of carbon-containing products on the inner walls of the unit. These deposits, which consist of coke or heavy pyrolysis tars that are condensed and more or less agglomerated, limit heat transfer in the cracking zone (in a pyrolysis pipe coil) and the indirect quenching zone (effluent quenching exchanger), thus requiring frequent shutdowns to decoke the unit.
The standard cycle periods (operation between two complete chemical decokings of the cracking zone, with air and/or with vapor) are either set (scheduled shutdowns) or variable, depending on the coking of the unit, and generally are spread between 3 weeks and 12 weeks for feedstocks such as naphtha and liquefied petroleum gases.
It is known to one skilled in the art that the coking problems that are encountered during the cracking of heavy feedstocks (atmospheric gas oils, heavy gas oils, distillates under vacuum) are much more serious than those that are encountered on standard feedstocks, such as naphtha.
Consequently, these feedstocks cannot be cracked in standard steam-crackers that are designed for cracking naphtha, and cannot be cracked according to the known processes that in special furnaces typically comprise direct quenching (with pyrolysis oil) of the steam-cracking effluents; this considerably impairs the energy balance of the unit (no high-pressure vapor production).
The known processes that make it possible to have flexibility with respect to the heavy feedstocks are therefore incompatible with the existing steam-cracking units on standard feedstocks, and they have a greatly degraded energy balance.
Furthermore, in the case of relatively light feedstocks, there is a tendency also to use as steam-cracking feedstocks poor-quality naphthas such as C.sub.4, C.sub.5 recycling fractions, thermal cracking gasolines, or the core of fraction of catalytic cracking (FCC). These olefinic feedstocks lead to significant coking problems, in particular with a high degree of severity.
Furthermore, the applicants have already proposed (EP-A-419 643, EP-A-425 633 and EP-A 447 527) a process for in-service decoking of steam-cracking units, by injecting erosive solid particles in order to solve coking problems and to obtain continuous or approximately continuous steam cracking (for example, cycle periods on the order of 1 year).
For a given feedstock, this process consists in allowing a layer of coke to form and mature on the inner walls of the cracking coil and then injecting erosive particles (for example, mineral particles that are hard, with a diameter of less than 150 micrometers, spherical or angular) in sufficient quantity to stabilize approximately the coking state of the pipes without totally eliminating the pre-layer of coke, which plays a protective role for these pipes.
This process requires good knowledge of the coking speeds of the feedstock that is being considered and a coil design such that there is a certain correspondence between the local coking speeds, which are related to the progress of cracking along the coil, and the intensity of erosion, which is related to the speed profile along the coil and to the nature of the erosive particles. With, on the one hand, simulations of coking speeds and the profile of circulation speeds in the coil and, on the other, pilot experiments, it is possible to provide for approximately continuous steam-cracking conditions of the feedstock under study.
It can be ensured that there is very little if any erosion of the pipes, and erosion can be monitored by the analysis of metal traces (iron, chromium, nickel) in the powders that are recovered.
The applicants have therefore sought to improve this process, which can be applied to the cracking of a given feedstock, in the case of a flexible furnace that can successively process a large number of different feedstocks, with variable operating conditions (flow rate, degree of dilution, degree of severity of cracking).
Pilot tests have been carried out and have provided several unexpected results:
It was actually found that the initial coking of the coil (at the beginning of the cycle) could vary to a very significant extent depending on the feedstock, including for feedstocks that are slightly different in terms of chemical composition but are of different origins. It has not been possible to explain this completely, and it may be due to impurities that are present in the feedstock.
Furthermore, the effectiveness of decoking has proven to depend in particular on the feedstocks and operating conditions (different nature of the coke). In particular, it was found that at the beginning of the reaction zone the light feedstocks: C.sub.3, C.sub.4, light naphtha, produce a much more fragile catalytic coke (by 5 to 10 times) than the asymptotic coke that predominates in the middle and at the end of the reaction zone. For these feedstocks it is therefore desirable to limit the circulation speed in this zone to maintain a protective coke layer and/or to avoid risks of erosion of the cracking pipes.
Thus, it has not been possible to predetermine the quantities of particles that are suitable for each feedstock and each operating condition without preliminary tests, and such tests cannot be carried out in the case of a flexible industrial furnace. In addition, with regard to the prevention of erosion risks, the geometry of the cracking reactor that is suitable for a given feedstock is not the same as that which is suitable for another feedstock that has a degree of dilution and a type of coke that are different, for which the appropriate profile of circulation speeds will be different.
Finally, because of the difficulties in obtaining reliable and specific measurements of skin temperatures of the pipes by optical pyrometry, as well as due to fluctuations of these temperatures and the loss of load under the variable operating conditions, it is very difficult to monitor effectively the state of coking of the pipe without resorting frequently to a constant reference state, which cannot be done for a flexible industrial furnace, and thus to be able to control in real time the coking of a pyrolysis coil.
The process has therefore proven difficult to use industrially under variable operating conditions, and it has not been possible to keep any trace of erosion from arising in cracking pipes for all the pilot tests.
It thus appeared that the continuous steam-cracking process could not be adapted in the case of a flexible furnace and should be reserved solely for cracking of identical or similar feedstocks under relatively stable conditions.
Furthermore, the applicants have noted that elimination of deposits in the indirect quenching exchanger could be achieved much more easily than in the pyrolysis pipes and that, even in the case of injection of particles in excess quantity, no erosion was noted.
Thus, it appeared, surprisingly enough, that the carbon-containing deposits of the quenching exchanger, in particular in the case of heavy feedstocks, were much more fragile than the coke of the cracking pipes. It was found, actually, that the brittleness, compared to the erosion by the solid particles under test, was at least 25 times greater for the coke of the quenching exchanger than for the asymptotic coke of the pyrolysis pipes.
The absence of erosion that is noted for the pipes of the exchanger themselves is explained by the fact that the circulation speed of the particles is much lower in the quenching exchanger than in the pyrolysis pipes and that their temperature is very low (approximately 330.degree. C. as opposed to typically 1000 to 1100.degree. C. for the pyrolysis coil). In addition, the quenching exchanger pipes are straight, without bends, which eliminates the risks of point erosion.
Furthermore, it has appeared that over a long period of time the coking speeds in the cracking pipes remain on the same order of magnitude for the heavy feedstocks (for example, gas oil) as for light feedstocks and that the true bottleneck for flexibility with respect to the heavy feedstocks lies in the excessive clogging of the indirect quenching exchanger. Thus, it has been found, in a non-obvious way, that existing naphtha steam-cracking units were also able to crack heavy feedstocks such as gas oils and distillates of suitable quality under a vacuum or olefinic feedstocks with a high degree of severity if the quick clogging of the indirect quenching exchangers could be prevented.
Likewise, for light olefinic feedstocks, such as C.sub.4, C.sub.5 recycling fractions or gasoline fractions of catalytic cracking with a low octane number, it was also found that the clogging of quenching exchangers was much more severe for the quenching exchangers, in particular with a high degree of severity.
Therefore a new process has been proposed, in French Patent Application No. 94/15743, a new process for flexible steam cracking that is compatible with the existing steam-cracking units which make it possible to treat various feedstocks according to variable operating conditions, without degrading the thermal balance of the units, without significant risks of erosion, and at a modest investment cost.
One of the main characteristics of this process is that the bulk of the erosive solid particles (at least 70% and up to 100%) are injected downstream from the pyrolysis pipes and upstream from the indirect quenching exchangers.
This makes it possible to limit or eliminate the technological risks of erosion of the pyrolysis pipes by minimizing or preventing the circulation of particles in these pipes, while ensuring flexible operation thanks to a decoking effect on the part of the quenching exchangers.
The application of this process to a type of modern furnaces with very short dwell times (0.07 to 0.27 second) faces technical obstacles, however: these furnaces actually contain a very large number of small-diameter pyrolysis pipes that are connected downstream to a large number of primary quenching exchangers, generally of the double-pipe type, straight or U-shaped, which are connected downstream to a very small number (generally 1 or 2) of secondary quenching exchangers, before the transfer of cooled cracked gases to the direct downstream quenching.
The application of the process according to French Patent Application No. 9415743 therefore requires that particles be injected upstream from each of the primary quenching exchangers, of which there are typically a large number (for example, 20 per furnace) and with higher input temperatures (850 to 900.degree. C.) than in the standard steam-cracking furnaces.
The circulation speeds in these primary quenching exchangers are also very high (generally, considerably greater than 100 m/s), which increases the risks of erosion, in particular in the output bends, of which there are many.
The application of the flexible steam-cracking process is therefore complex and difficult for this type of furnace.